Mechanics of Advanced Materials and Structures

ISSN: 1537-6494 (Print) 1537-6532 (Online) Journal homepage: http://www.tandfonline.com/loi/umcm20

Parametric study for concrete cover separation

failure of retrofitted SNSM strengthened RC beams

Ahmad Azim Shukri, Mohd Fazaulnizam Shamsudin, Zainah Ibrahim & U.

Johnson Alengaram

To cite this article: Ahmad Azim Shukri, Mohd Fazaulnizam Shamsudin, Zainah Ibrahim
& U. Johnson Alengaram (2018): Parametric study for concrete cover separation failure of
retrofitted SNSM strengthened RC beams, Mechanics of Advanced Materials and Structures, DOI:
10.1080/15376494.2018.1482034

To link to this article: https://doi.org/10.1080/15376494.2018.1482034

Published online: 11 Jun 2018.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=umcm20
MECHANICS OF ADVANCED MATERIALS AND STRUCTURES
, VOL. , NO. , –
https://doi.org/./..

ORIGINAL ARTICLE

Parametric study for concrete cover separation failure of retrofitted SNSM

strengthened RC beams
Ahmad Azim Shukria , Mohd Fazaulnizam Shamsudinb , Zainah Ibrahima , and U. Johnson Alengarama
a
Faculty of Engineering, University of Malaya, Kuala Lumpur, Malaysia; b Department of Civil Engineering, University of Nottingham,
Nottingham, UK

ABSTRACT ARTICLE HISTORY

The side near-surface mounted (SNSM) method is a new flexural strengthening method for reinforced con- Received  November 
crete (RC) beams which was proposed to allow near-surface mounted (NSM) strengthening to be applied on Accepted  May 
beams with small width. As a relatively new strengthening method, further studies are needed to determine KEYWORDS
the effects of strengthening parameters on the flexural performance of RC beams. In response to that, this Near-surface mounted;
paper presents a parametric study on the concrete cover separation failure of SNSM strengthened beams numerical analysis; partial
using a simulation method based on the moment-rotation (M/θ) approach. interaction; reinforced
concrete; side-NSM

1. Introduction
Reinforced concrete (RC) structures tend to face some degree of
strength loss due to aging. To restore the lost strength, or even
increase the structural strength beyond the original strength,
structural strengthening [1]–[3] can be applied. Among the
newer strengthening method is near-surface mounted (NSM)
method [4]–[13]. The NSM method consists of NSM reinforce-
ments, which is usually some form of fiber reinforced polymers
(FRP) that is placed within a groove made at the soffit of RC
beams to strengthen it in flexure.
Among the problems with applying the NSM method is that it
requires the RC beam to be considerably wide. A closely spaced
arrangement of NSM bars will cause an overlap of stresses, which
causes the tensile stress at the concrete-epoxy interface to be
magnified and cause concrete split failure [14]. The ACI 440
guideline, based on the research work of [15] states that the min-
imum clear groove spacing for NSM bars should be greater than
twice the depth of the groove to avoid the overlapping of stresses,
while the edge distance should be four times the depth of the
groove to minimize edge effects. To make the NSM method
applicable to beams with small width, a minor modification to
the NSM method was introduced. The modified method, named
side-NSM (SNSM) method changes the location of the NSM
reinforcement from the soffit of the RC beam to the side of the
beam at the same level as the tension reinforcement. Addition-
ally, the SNSM method allows strengthening to be applied on
beams with walls beneath them [16].
The SNSM method is a relatively new method; there are
very limited research that has been done on the SNSM method
thus far. Hosen et al. [17] used FRP and steel bars as SNSM
reinforcements; it was reported that the FRP bars gives a higher
ultimate load but lower ductility while the steel bars is the oppo-
site with lower ultimate load but higher ductility. The newest
study on SNSM was done by Sharaky et al. [16] who performed
experimental study on GFRP SNSM strengthened beams, where
the GFRP bars have ends that are bent into different degrees of
inclinations. The bent end conditions for SNSM method was
found to be less beneficial compared to bottom NSM, which
have good confinement due to the bent ends being encased in
concrete.
Shukri et al. [18] studied the behavior of SNSM strengthened
RC beams that has been applied a precracking load. The pre-
cracking loads were used to induce flexural cracks on beams
prior to strengthening them with the SNSM method. The pur-
pose was to simulate the condition of actual beams which would
have flexural cracks due to service load prior to being retrofitted.
For the sake of brevity, beams that are applied precracking load
prior to being strengthened will be referred to as retrofitted
beams for the rest of this paper. Shukri et al. [18] reported that
SNSM retrofitted beams have a slight decrease in ultimate load
of up to 3.3% compared to virgin SNSM strengthened beams.
Importantly, it was found that the SNSM retrofitted beams have a
higher flexural stiffness than SNSM strengthened beams without
pre-existing flexural cracks. To study this occurrence, Shukri et
al. [18] performed an analysis using the moment rotation (M/θ )
approach; further detail on the M/θ will be given later. From the
analysis, it was found that the retrofitted beams have a longer
crack spacing compared to virgin strengthened beams due to
the larger concrete area adjacent to the steel reinforcement when
the flexural cracks formed. The longer crack spacing causes the
resulting curvature to be smaller compared to virgin strength-
ened beams at the same value of moment.
One important characteristic of the SNSM method that can
be determined from the experimental works that has been done
so far is that concrete cover separation is the sole mode of pre-
mature failure. Premature failures refer to failure modes for

CONTACT Zainah Ibrahim zainah@um.edu.my Department of Civil Engineering, Faculty of Engineering, University of Malaya, Kuala Lumpur , Malaysia.
Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/umcm.
©  Taylor & Francis Group, LLC
2 A. A. SHUKRI ET AL.
strengthened RC beams that occur prior to the rupture of the
strengthening reinforcements. The concrete cover separation
[19]–[21], also called end debonding or end cover separation,
is the failure mode commonly reported in experimental works
on NSM strengthened RC beams. The debonding crack forms
near the curtailment location of NSM reinforcement, which
then propagates upwards until it reaches the shear link of the
NSM strengthened beam. The crack then propagates horizon-
tally, causing the NSM reinforcement to be debonded along with
the concrete cover of the beam.
It is clear that there need to be a study done on the concrete
cover separation failure of SNSM strengthened RC beams. Addi-
tionally, the parametric study should involve SNSM retrofitted
RC beams so that the study is relatable to actual beams. To
this end, this paper aims to study the effect of several selected
parameters on the concrete cover separation failure of retrofitted
and virgin SNSM strengthened RC beams by using the M/θ
approach. The M/θ approach [18], [22]–[36] is a relatively new
simulation method. The main characteristic of this approach is
the application of the partial interaction theory [37]–[39], which
to summarize states that where a tensile crack intercepts a rein-
forcing bar, infinite strains are theoretically induced in the rein-
forcing bar that must be relieved by a slip between the steel rein-
forcement and the concrete. By applying a numerical solution
to simulate the slip of steel reinforcement, various mechanics
of RC beams, such as the formation of flexural cracks, widen-
ing of flexural cracks and tension stiffening can be accounted
for. The advantage that the M/θ approach has over conventional
moment-curvature (M/χ ) approach is the fact that it can readily
simulate these mechanics without resorting to empirical formu-
lations, such as the use of Branson’s equation to simulate tension
stiffening. It should be noted that while the M/θ approach can
directly simulate mechanics of RC beams, empirical formula-
tions in terms of material models are still needed, such as stress–
strain models and bond stress-slip models.
Previously Shukri et al. [18] extended the M/θ approach to
simulate the behavior of SNSM strengthened RC beams. The
simulated load–deflection curve could follow the general shape
of the experimental load–deflection curve well. The proposed
method however was not suitable for parametric study as it
could not simulate concrete cover separation, which is a type of
debonding found to occur on SNSM strengthened beams tested.
This paper will further extend the existing M/θ approach to
allow it to simulate concrete cover separation of SNSM strength-
ened RC beams. In the initial sections of this paper, the funda-
mental mechanics of the M/θ approach will first be presented. A
numerical tension stiffening simulation will then be presented,
followed by the M/θ simulation. The method used to simulate
the effect of concrete cover separation on SNSM strengthened
RC beams will then be shown, along with the difference involved
when applying the method on virgin and retrofitted beams. The
proposed method is then validated against published experi-
mental results. Parametric study would then be done on several
selected parameters.
2. Tension stiffening simulation
In undisturbed regions of an SNSM strengthened RC beam,
which remains without any flexural cracks, there exists perfect
bonding between the steel and SNSM reinforcements and the
concrete adjacent to them, where there is strain compatibility
between the reinforcements and the adjacent concrete and no
slip of reinforcement occurs. Once flexural cracks form, par-
tial interaction causes both the steel and SNSM reinforcement
begins to slip from the concrete. Consider the tension stiffening
prism in Figure 1(b), which is made up of the steel reinforcement
and the adjacent concrete. The load applied on the beam causes
the pullout force Pr , which in turn causes the steel reinforcement
and SNSM reinforcement to slip by an amount δ r ; the steel rein-
forcement and SNSM reinforcement will have the same amount
of slip as both have the same height from beam soffit. Due to the
bond that exists between steel/SNSM reinforcement and con-
crete, the pullout force is gradually transferred from and onto
the adjacent concrete. As the pullout force is reduced, the slip
also gradually drops further away from the crack face as shown
in Figure 1(c).
As the load applied on the beam is increased, the amount
of load transferred to the concrete will also become higher;
this causes the steel reinforcement strain, εr to decrease while
the concrete strain, εc increases as shown in Figure 1(d) and
(e), respectively. Once the concrete strain reaches the concrete
cracking strain, εcr , a primary crack will form. The mechanism
described here will continue until there are primary cracks along
the length of the beam with a crack spacing of Scr as shown
in Figure 1(a); the value of Scr here represents the minimum
crack spacing, although when moment gradient is present on
the beam, which tends to be the case, it is more likely and con-
servative to take the spacing as Scr [24]. With the formation of
primary cracks, the loading of each RC beam segment becomes
symmetric, as shown in Figure 1(f). As such only the half-length
of the RC section, Ldef , needs to be considered [24]. The distri-
bution of slip within beam section of length Scr is as shown in
Figure 1(g), while the strain of reinforcement and strain of con-
crete is shown in Figure 1(h) and (i), respectively; the symmetry
seen in these three figures reflect the symmetry of forces acting
on the beam section.
To consider the effects of concrete cover separation on SNSM
strengthened beams, consider the comparison of tension stiffen-
ing prism size in the strengthened section of the beam and the
debonded section of the beam is given in Figure 2(b) and (c)
respectively. When concrete cover separation occurs, as shown
in Figure 2(a), the SNSM reinforcement in the debonded sec-
tion no longer contributes to the beam and only the tensile steel
reinforcement needs to be considered. Since the SNSM is applied
on the side of the beam, the tension stiffening prism of the ten-
sile steel reinforcement is not affected, as shown in Figure 2(c).
This contrasts with NSM strengthened beams, where the steel
reinforcements will have a reduced tension stiffening prism size
when concrete cover separation occurs [32].
For analysis purpose, the tension stiffening prism’s depth can
be taken as 2c, where c is the distance from beam soffit to center
of the steel reinforcement [29], [40]. The width of the prism for
SNSM reinforcement is taken as the groove width, while for steel
reinforcement the width is the sum of the diameters of steel rein-
forcement and shear link and two times of any remaining width
of the concrete cover that is not part of the SNSM groove.
A numerical tension stiffening simulation based on the par-
tial interaction theory has been applied by many researchers
to simulate the slip of steel reinforcement [32], [35], [37]–
[39], which will serve as the basis for the numerical procedure

Figure . Formation of primary cracks: (a) SNSM strengthened RC beam; (b) tension stiffening prism with variable length; (c) distribution of reinforcement slip; (d) distribu-
tion of reinforcement strain; (e) distribution of concrete strain; (f) tension stiffening prism of length Scr ; (g) distribution of reinforcement slip; (h) distribution of reinforcement
strain; and (i) distribution of concrete strain.
presented here. The numerical analysis firstly determines the
length of primary crack spacing, Scr . The analysis is then reduced
to half the crack spacing, Ldef which ends with a load-slip rela-
tionship for the steel reinforcement and the SNSM reinforce-
ment each. It should be noted that the numerical procedure is
the same for either SNSM or steel reinforcement, but the crack
spacing Scr is controlled by the steel reinforcement; hence the
numerical tensions stiffening simulation should be performed
on the steel reinforcement first so that Scr and Ldef can be deter-
mined. The numerical procedure is similar to the one used by
Shukri and Jumaat [32]; the procedure is reproduced here for
readers’ easy reference and is explained below:
1. The tension stiffening prism is made up of concrete area
Ac and steel reinforcement area Ar . The prism is then
divided into elements of length Ls ; in this paper the Ls
is taken to be 0.1 mm, which is small enough such that
the stresses and strains acting along each element can be
considered constant.
2. At the location of crack, an initial value of slip, δ(1) is
set. The load acting on the concrete, Pc (1) is zero as the
concrete-concrete interface is not touching. The load act-
ing on the steel reinforcement, Pr (1) is assumed.
MECHANICS OF ADVANCED MATERIALS AND STRUCTURES 3
3. The bond force acting on the steel reinforcement is B(1)
= τ (1)Lper Ls , where τ (1) is determined using any proper
bond stress-slip model while Lper is the circumferen-
tial perimeter of the reinforcement. The strain of steel
reinforcement is εr = Pr (1)Ar /Er where Er is the rein-
forcement’s elastic modulus. The change in slip from this
prism element to the next is ࢞δ(1) = (εr (1) – εc (1))Ls .
4. The slip in the next prism element is δ(2) = δ(1)
– ࢞δ(1). The forces acting on the reinforcement
and concrete are Pr (2) = Pr (1)–B(1) and Pc (2) =
Pc (1)+B(1), respectively. The concrete strain is εc =
Pc (i+1)Ac /Ec where Ec is the concrete modulus. This
procedure is repeated for the next prism element as
well.
5. Steps 2–4 are repeated with different values of assumed
Pr (1) until the initial slip δ(1) can be reduced to 1% of
its value by δ(n), where n refers to number of prism ele-
ments required
6. Steps 2–5 are continued until εc ࣙ εct , where εct is the
concrete cracking strain. The primary crack length Scr
can be determined from the total length of prism ele-
ments from δ(1) to δ(n).
4 A. A. SHUKRI ET AL.
Figure . Concrete cover separation on SNSM strengthened RC beams: (a) side
view of debonded SNSM strengthened beam; (b) SNSM strengthened beam cross-
section prior to debonding; and (c) SNSM strengthened beam cross-section prior to
debonding.
7. With the formation of primary crack, steps 1–5 are then
repeated with total prism element length n limited to
Ldef = Scr /2. The steps are repeated and the values are
recorded until a load-slip (Pr /δ) relationship for the steel
reinforcement is obtained.
8. Step 7 is then repeated with the SNSM reinforcement to
obtain the (Pr /δ) relationship for the SNSM reinforce-
ment. Note that Scr is controlled by steel reinforcement
and not the SNSM reinforcement, so procedure 6 does
not need to be repeated with the SNSM reinforcement.
3. Moment-rotation simulation
With the load-slip relationship of the steel reinforcement and
SNSM reinforcement determined, the M/θ simulation can now
be performed. Consider Figure 3, where a beam section of length
Ldef is rotated by θ degree due to moment M. While the RC beam
is uncracked, the forces that causes deformation on the beam as
shown in Figure 3(a) is to be determined using the stress–strain
relationships of each material. The depth of neutral axis dna is
then adjusted until equilibrium of forces is achieved; the actual
value of moment M which causes rotation θ is then determined.
To take into account the formation of concrete wedges and the
resulting concrete softening, the stress–strain model for con-
crete presented by Popovics [41] was used. As the stress–strain
of concrete is size dependent, the concrete stress–strain is com-
monly adjusted for size [22, 33, 42–44] with the size-dependent
stress–strain method by Chen et al. [42].
When flexural cracking occurs, slip of reinforcements
occur such that the strains of both the SNSM reinforcements
and steel reinforcements are no longer constant along length
Ldef . The forces acting on the steel reinforcement and SNSM
reinforcements is thus determined using the load-slip relation-
ship obtained from the tension stiffening analysis, where the slip
δ r is determined from the deformation profile in Figure 3(a).
Note that the value of slip δ r applies to both SNSM reinforce-
ments and steel reinforcements, as both have the same value of
slip. The dna was again adjusted until equilibrium of forces was
obtained, after which the value of M was determined. The whole
process was repeated for different values of θ in order to obtain
a M/θ relationship. From there to obtain the M/χ relationship is
only a matter of dividing the θ with Ldef . Two types of M/χ are
needed to simulate the whole SNSM strengthened RC beam:
1. The moment-curvature of the strengthened section of
the beam, (M/χ )s
2. The moment-curvature of the debonded/unstrengt-
hened section of the beam, (M/χ )u
4. Determining load–deflection relationship and
simulating concrete cover separation
The procedure to determine the load–deflection of the SNSM
strengthened beam will be presented here. To take into account
the occurrence of concrete cover separation failure, the global
energy balance approach (GEBA) [45–48] will be used in con-
junction with the M/θ approach. The fundamental principle for
GEBA is that it assumes that debonding cracks will always occur
on strengthened RC beams; it is then only a matter of deter-
mining whether there is enough strain energy for this debond-
ing crack to propagate to cause failure for the SNSM strength-
ened RC beam. The bending strain energy can be obtained using
the M/χ of the strengthened and debonded state of beam at
the location of SNSM curtailment as shown in Figure 4. With
assumption that moment remains constant due to sudden pro-
cess in debonding propagation, the strain energy released during
debonding is thus the area Wa in Figure 4.
From the value of Wa , the energy release rate (Ga ) is then
determined as:
Wa
Ga = (1)
b × L
where b is the width of debonding crack, which is taken as two
times the width of the SNSM groove and ࢞L is the change in
debonding crack length, which is taken as 1 mm for gradual
propagation of the debonding crack. The value of Ga is then
compared against the fracture strength of concrete, Gmax . When
Ga > Gmax , the debonding crack will progress toward the center
of the beam. The load at which this occurs will be referred to as
the debonding load, Pd in this manuscript. However this does
not mean that the beam has completely failed but rather that the
debonding crack has progressed by the length ࢞L . Shukri and
Jumaat [32] proposed a procedure for gradual propagation of
the debonding crack, which will be used in this manuscript.
A value of load FP is set and a debonded length, Ld is assumed
to have already formed on at the location of curtailment. When
applying the GEBA on beams strengthened with FRP sheet [48]
or NSM [32] it is assumed that the shear crack that causes the
debonding propagates at an angle of 45° to the beam axis until it
reaches the shear link, which means the length of Ld is equal to
the concrete cover’s depth as shown in Figure 5(a) and Figure
5(b). This assumption does not apply to SNSM strengthened

Figure . Moment-rotation analysis: (a) beam segment and deformation profile; (b) strain profile; (c) stress profile; and (d) force profile.
Figure . Difference in M/χ at location of SNSM curtailment before and after prop-
agation of debonding crack.
beams since the shear link is not above the SNSM reinforcement,
but at its side. From the experimental studies on SNSM strength-
ened beams [16–18], it was noted that the shear crack starts to
propagate horizontally as it reaches the SNSM reinforcement,
such that Ld = 0 as shown in Figure 5(c).
Figure . Initial debonded length for different strengthening methods: (a) FRP
sheet or plate; (b) NSM reinforcement; and (c) SNSM reinforcement.
MECHANICS OF ADVANCED MATERIALS AND STRUCTURES 5
In this paper, the change in debonding crack length, ࢞L is
taken as 1 mm in the beginning. An initial load, Fa is applied
on the simulated beam. The applied moment, Ma is then deter-
mined and the commonly used double integration method is
then used to determine the deflection by using (M/χ )s , and
(M/χ )u for the strengthened and nonstrengthened sections of
the beam respectively. The value of Ga at the end of the location
of SNSM curtailment is then determined; if Ga > Gmax , the ࢞L is
increased by 1 mm. The value of Ga is then calculated again, and
this process is repeated until Ga < Gmax . The debonded length
Ld is then increased by ࢞L and the applied load Fa is increased
and the whole procedure is repeated. At some value of Ma it will
no longer possible to determine the value of deflection as the
applied moment Ma is beyond the range of moment in (M/χ )u .
The failure load has then been reached and the beam has expe-
rienced debonding failure. Alternatively, if this does not occur
and the applied Ma exceeds the range of (M/χ )s instead, then
the beam does not fail by concrete cover separation. The proce-
dure to obtain the load–deflection relationship of SNSM beam
discussed here is also shown as a flowchart in Figure 6.
5. Simulating SNSM retrofitted RC beams
The differences involved when simulating SNSM retrofitted RC
beam will be presented here. To reiterate, retrofitted RC beams
refer to RC beams that have been in service prior to being
strengthened; a strengthened virgin beam on the other hand is
not. While most lab work focuses on strengthened virgin beams,
their results may not be identical to real world results. When a
beam needs to be retrofitted or strengthened, there will already
be primary cracks already present on the beams. The primary
crack spacing of these beams, Scr-s , is likely to be smaller than
the primary crack spacing of strengthened virgin beams, Scr-v as
shown in Figure 7.
For SNSM strengthened beams, Shukri et al. [18] showed
that the difference in crack spacing is due to the change in
tension stiffening prism size. Prior to being retrofitted, the
tension stiffening prism is taken to be 2 c times the width of
the beam divided between the number of steel reinforcements.
The tension stiffening prism size is thus larger compared to the
size for strengthened RC beam, as shown in Figure 7(a). When
the beam is retrofitted, the beam will have the tension stiffening
prism size reduced, but the primary cracks will have occurred
and the crack spacing will be Scr-v . The combination of a smaller
tension stiffening prism and longer crack spacing causes the
curvature of a retrofitted beam to be smaller compared to the
6 A. A. SHUKRI ET AL.

Insert beam geometric Table . Details of SNSM strengthened RC beams.

Start properties and (M/θ)s and
Beam designation dn (mm) Pcr (kN)
(M/θ)u relationships
SNC  —
SNC  —
Divide θ by Ldef to obtain (M/χ)s and (M/χ)u relationships SNC  —
PSNC  .
PSNC  
PSNC  .
Apply load Fa and determine the resulting Ma
Note: Msnsm = material for SNSM reinforcement; dn = diameter of
SNSM reinforcement; Pcr = precracking load.

∆L=1mm
Increase Fa 6. Validation using experimental results
Determine maximum beam deflection by The experimental results by Shukri et al. [18] will be used to val-
double integration method using (M/χ)s and idate the proposed method. The load versus mid-span deflec-
(M/χ)u relationships tion results of six SNSM strengthened RC beams are available,
with three of the beams applied precracking loads (Pcr ) of 22.5,
30, and 37.5 kN each. The purpose of these precracking loads
is only to induce flexural cracks in order to simulate the condi-
Is it possible to obtain tion of retrofitted beams as seen in practice. The different values
No End
deflection? of precracking loads correspond to the different flexural rigid-
ity of the beams due to the different sizes of steel reinforcement
used. After the precracking load was applied, the beams were
Yes unloaded, strengthened using SNSM method and applied load
again up to failure. All the beams used carbon FRP (CFRP) bars
Record Fa Determine G a
and the as SNSM reinforcement. The SNSM CFRP reinforcements had
resulting a diameter of either 8, 10, or 12 mm. The further details on the
deflection of beams are available in Table 1. In Table 2, the material properties
beam.
Is Ga>Gmax? Yes ∆L=∆L + 1mm
of the concrete, steel reinforcement and FRP bars are given.
The beams have a dimension of 125 mm × 250 mm with a
clear cover of 27 mm; the length of the beams was 2300 with
No
2000 mm as the effective span and a shear span of 650 mm. The
Ld=Ld+∆L SNSM reinforcements have a bonded length of 1900 mm. Two
steel reinforcements with 12 mm diameter were used as tensile
Figure . Flowchart for determining the load–deflection of NSM strengthened reinforcement. Two steel reinforcements with 10 mm diameter
beams.
deformed bars were used as compression reinforcement up to
the shear span zone. Shear reinforcement was provided through
curvature of a virgin beam for a similar value of moment. This 6 mm diameter bars, distributed along the length of the speci-
results in the flexural stiffness of retrofitted beams to be higher mens except in the constant moment region to prevent it from
than strengthened virgin beams. influencing crack propagation.
While a retrofitted beam may have some existing primary
cracks, the low moment regions of the beam may still be undis-
turbed, as shown in Figure 7(c). If load is applied such that 7. Material models
primary cracks appear on these low moment regions after Several material models were used in this study. It should be
being retrofitted, the crack spacing will be Scr-v corresponding noted that the models used in the M/θ approach are intended
to the crack spacing for the smaller tension stiffening prism to act as input for the analysis, they may be refined or changed
for strengthened RC beam as shown in Figure 7(b). For sim- in order to produce results that are more accurate [23]. Only a
plicity, the behavior of an SNSM retrofitted beam can sim- general information about the material models are given here to
ply be determined by using the crack spacing of Scr-s for the keep the paper short; more information about the models can be
entirety of the beam with good correlation between simulated obtained using the reference given.
and experimental results [18]. When applying GEBA on SNSM For the steel reinforcements a bilinear stress–strain model
retrofitted beam, however, it is more conservative and accu- with strain hardening was used, the CFRP bars used a linear
rate to use Scr-v for the regions near the end of the SNSM
reinforcement. Table . Material properties.
To summarize, GEBA procedure for SNSM retrofitted beam
fc (MPa) Ey (MPa) σ y (MPa) σ u (MPa) Ef (MPa) σ f (MPa)
as discussed in the previous section is still applicable here, with
one change: when calculating the Ga , the rotation, θ for (M/θ )s  ,   , 
and (M/θ )u are divided by Ldef-v instead of Ldef-s to obtain Note: fc = concrete strength (cylinder); Ey = steel elastic modulus; σ y = steel yield
(M/χ )s and (M/χ )u , respectively. Ldef-v and Ldef-s are half of the strength;; σ u = steel ultimate strength; Ef = FRP modulus; σ f = FRP tensile
primary crack spacing Scr-v and Scr-s , respectively. strength.

Figure . Comparison of primary crack spacing: (a) Virgin RC beam; (b) Virgin SNSM strengthened RC beam; and (c) SNSM retrofitted RC beam.
stress–strain model. The bond-slip model by CEB-FIP [49] for
the steel reinforcement was used in the tension stiffening simu-
lation. For the SNSM reinforcements, the bond-slip model by
De Lorenzis [50] which was derived for NSM reinforcements
is used. The maximum bond stress, τ max was obtained using
the bond strength model by Hassan and Rizkalla [14]. Popovics’
[41] concrete compressive stress–strain model was used. The
concrete stress–strain model was adjusted for size using the
size-dependent stress–strain method proposed by Chen et
al. [42].
8. Comparison against published experimental
results
The comparison of simulated and experimental load versus mid
deflection curves for beams SNC8, SNC10, PSNC8, and PSNC10
are given in Figure 8. All the beams were reported to have
failed by FRP rupture instead of concrete cover separation. Their
inclusion in this study is to determine whether the method pro-
posed in this paper can accurately simulate their behavior. The
simulated curve was able to follow the shape of the experimental
curve well. The simulated failure loads and deflections at failure
are also adequately accurate.
The comparison of load versus mid-span deflection for
beams SNC12 and PSNC12 is shown in Figure 9. The beams
were reported to have failed by concrete cover separation. From
Figure 9, it can be seen that the simulated curve was able to pre-
dict the failure load and failure deflection accurately. Figure 9
also includes the simulated curves without GEBA, which means
that these curves could not simulate concrete cover separation
failure. It can be seen that without the use of GEBA, the simu-
lated result overpredicts the failure load of the beams by a con-
siderable degree.
MECHANICS OF ADVANCED MATERIALS AND STRUCTURES 7
9. Parametric study
Using the proposed method, parametric studies were conducted
to determine the effects of several parameters on the overall
behavior of SNSM strengthened RC beams. The detail for the
parametric study is given in Table 3. The parameters tested
are concrete strength (fc ), elastic modulus of SNSM reinforce-
ment (Er-snsm ), bond strength of epoxy adhesive (τ max ), and
location of SNSM curtailment which is given in term of the
distance of the SNSM reinforcement’s end to the beam’s sup-
port (La ). This parametric study also includes comparison of
results between retrofitted and virgin SNSM strengthened RC
beams in order to study the difference in behavior between the
two beam conditions. The geometric and material properties
of beam SNC12 and PSNC12 are used as the reference for this
parametric study, where beam SNC12 represents virgin SNSM
strengthened beam and PSNC12 represents SNSM retrofitted
beams.
The result of the parametric study is given in Table 3. The
debond load, Pd is the load at which the energy release rate Ga
is found to be greater than Gmax , causing the debonding crack
to start propagating horizontally toward the center of the beam.
This does not mean the beam would immediately fail, as the
debonding crack’s propagation can be a gradual process. As can
be seen in Table 3, for most beams there are considerable differ-
ence in the values of Pd before the beams finally fail at the failure
load, Pf . Nearly all the beams failed due to concrete cover separa-
tion, although it was found that beams with low values of Er-snsm
and τ max-snsm failed due to concrete crushing instead. Also pre-
sented in Table 3 are the pre-yield stiffness, Ke of the beams,
which were determined by calculating the slope of the load–
deflection curve in the elastic region. The retrofitted beams con-
sistently have a higher Ke than the virgin beams, which agrees
8 A. A. SHUKRI ET AL.

Figure . Load versus mid-span deflection curves (a) SNC; (b) SNC; (c) PSNC; and (d) PSNC.

Table . List of parameters tested and summary of simulated results.

Parameter tested Beam condition Pd (N/mm ) Pf (N/mm ) Ke FM

fc =  N/mm Virgin . . . CCS

fc =  N/mm Virgin . . . CCS
fc =  N/mm Virgin . . . CCS
fc =  N/mm Retrofitted . . . CCS
fc =  N/mm Retrofitted . . . CCS
fc =  N/mm Retrofitted . . . CCS
Er-snsm =  kN/mm Virgin — . . CC
Er-snsm =  kN/mm Virgin . . . CCS
Er-snsm =  kN/mm Virgin . . . CCS
Er-snsm =  kN/mm Retrofitted — . . CC
Er-snsm =  kN/mm Retrofitted . . . CCS
Er-snsm =  kN/mm Retrofitted . . . CCS
τ max = . N/mm Virgin — . . CC
τ max = . N/mm Virgin . . . CCS
τ max = . N/mm Virgin . . . CCS
τ max = . N/mm Retrofitted . . . CC
τ max = . N/mm Retrofitted . . . CCS
τ max = . N/mm Retrofitted . . . CCS
La =  mm Virgin . . . CCS
La =  mm Virgin . . . CCS
La =  mm Virgin . . . CCS
La =  mm Retrofitted . . . CCS
La =  mm Retrofitted . . . CCS
La =  mm Retrofitted . . . CCS

Note: La = location of SNSM curtailment; Pd = debond load; Ke = effective pre-yield stiffness; Pf-n = normalizing failure load; FM = failure mode; CCS = concrete cover
separation; CC = concrete crushing.

Figure . Load versus mid-span deflection curves (a) SNC and (b) PSNC.
with that was reported by Shukri et al. [18] in their experimen-
tal study.
10. Discussion on the parametric study
The values of Pf given in Table 3 is shown in graph form in
Figure 10 to observe the trend of the values of Pf with respect
to the parameter tested. The simulated beams which failed by
concrete crushing are labelled with ‘cc’ in Figure 10, while unla-
beled ones are simulated to have failed by concrete cover sepa-
ration. The effect of fc on Pf is given in Figure 10(a). The values
of Pf were found to increase when fc is increased. The Pf for all
retrofitted beams were found to be lower compared to the virgin
beams by about 3–4%. At fc = 50 N/mm2 the effects of concrete
cover separation are less pronounced as the higher strength of
concrete increases the fracture strength, which in turn reduces
propagation of debonding crack. Where the debonding is less
pronounced, it can be seen that retrofitted beams have a higher
Pf than the virgin beam by about 1% due to the higher stiffness,
Ke of the retrofitted beam.
The effect of Er-snsm of SNSM reinforcement on Pf is given in
Figure 10(b). At Er-snsm of 62 kN/mm2 , it was found that both
the virgin and retrofitted beams failed through concrete crush-
ing rather than concrete cover separation. The low Er-snsm causes
MECHANICS OF ADVANCED MATERIALS AND STRUCTURES 9
the value of Wa to be smaller, which reduces the energy release
rate Ga , hence reducing the probability of concrete cover sepa-
ration. Increasing the Er-snsm to 124 kN/mm2 caused the beams
to fail by concrete cover separation, due to the larger Wa ; despite
this premature failure, it was found that the higher Er-snsm caused
the Pf to be much higher than beams with Er-snsm of 62 kN/mm2 .
Increasing the Er-snsm to 186 kN/mm2 caused the Pf to drop, as
the larger Wa caused the concrete cover separation to occur at
a lower debonding load, Pd , than the beams with Er-snsm of 124
kN/mm2 . A comparison of Pd is given in Table 3. From para-
metric study, it was found that the beams with Er-snsm of 124
kN/mm2 gave the highest Pf . It can also be observed from Figure
10(b) that when the beams do not fail by concrete cover sep-
aration, the retrofitted beams will have a higher Pf compared
to virgin beams. This attribute to higher flexural stiffness of
the retrofitted beams. However, when concrete cover separation
occurs as in the case when Er-snsm were 124 and 186 kN/mm2 , the
retrofitted beams will have a lower Pf compared to virgin beams.
The Pf of retrofitted beams was found to be 4% lower than vir-
gin beams; it was also noted that when Er-snsm was 62 kN/mm2 ,
the Pf of the retrofitted beam was 2% higher than the virgin
beam.
Figure 10(c) shows the effect of τ max on the Pf of SNSM
strengthened beams, where it can be seen that Pf decreases when
10 A. A. SHUKRI ET AL.

Figure . Simulated failure loads of beams in the parametric study (a) concrete strength; (b) elastic modulus of SNSM reinforcement; (c) bond strength of epoxy adhesive;
and (d) location of SNSM curtailment.

the τ max is increased. This is again related to the value of Wa 11. Conclusion
and Ga , which increases when higher values of τ max are used,
A method to simulate the behavior of SNSM strengthened RC
thus causing concrete cover separation to occur at a lower Pd
beams was presented; this method allows the simulation of con-
and reducing the Pf . It was found that when τ max was 5 N/mm2 ,
crete cover separation, which is the primary mode of premature
concrete cover separation did not occur and the Pf of virgin and
failure for SNSM strengthened beams. It is also shown how the
retrofitted beams were nearly identical. At higher τ max , where
concrete cover separation failure of SNSM retrofitted RC beams
concrete cover separation occurs, the Pf for retrofitted beams
can be simulated. Several conclusions can be made from the
are significantly lower than virgin beams. The difference in Pf
study:
between retrofitted and virgin beams was found to be about r The proposed method was validated and showed good
0–4%.
accuracy results using published experimental results.
The relationship between La and Pf is shown in Figure 10(d). r SNSM retrofitted strengthened beams was found to have
Pf was found to decrease with increasing La . A larger La results
approximately 3–4% lower failure load compared to virgin
in a larger moment acting around the end of the SNSM rein-
SNSM strengthened beams when concrete cover separa-
forcement; this causes a larger Wa and Ga , which then results
tion is a factor.
in a lower Pd and Pf . From Figure 10(d), it can be seen that r In cases where concrete cover separation failure did not
the retrofitted beams always have a lower Pf compared to virgin
occur or less pronounced, the failure load was found higher
beams. However, the difference in Pf due to beam condition was
in SNSM retrofitted beams by up to 1% due to approx-
less pronounced when compared to the difference in Pf caused
imately 15–19% higher flexural stiffness of retrofitted
by different values of La . The Pf for retrofitted beams were found
beams than virgin beams due to longer crack spacing of
to be 4% less than virgin beams.
the retrofitted beams.
From the result of Ke in Table 3, the retrofitted beams were r There is only a slight difference in failure load of SNSM
found to have a higher Ke compared to virgin beams in all cases
retrofitted beams compared to virgin SNSM strengthened
by about 15–19%. This finding is consistent with what was pre-
beams, although the small difference is negligible.
viously reported in the experimental study performed by Shukri r There is a considerable difference in the flexural stiffness of
et al. [18].
virgin and retrofitted beams that should not be neglected.

r It was found that retrofitted and virgin beam conditions
do not affect the failure mode of the SNSM strengthened
beams.
r While the current proposed method is complicated for
general design usage, it is recommended that this method
undergo further analysis using similar parametric studies
shown to produce simpler design procedures.
Conflict of Interest
The authors declare that they have no conflict of interest.
Funding
This study was supported by the University of Malaya, Grand Challenge -
SUS (Sustainability Science) Grant, project number GC003A-15SUS.
References
[1] R. A. Hawileh, M. Z. Naser, and J. A. Abdalla, Finite element simula-
tion of reinforced concrete beams externally strengthened with short-
length CFRP plates, Compos. Part B Eng., vol. 45, no. 1, pp. 1722–
1730, 2013. doi:10.1016/j.compositesb.2012.09.032.
[2] R. Hawileh, J. A. Abdalla, M. Z. Naser, and M. Tanarslan, Finite ele-
ment modeling of shear deficient RC beams strengthened with NSM
CFRP rods under cyclic loading, Am. Concr. Institute, ACI Spec.
Publ., vol. 2015–Janua, no. SP 301, pp. 69–85, 2015.
[3] R. A. Hawileh, H. A. Rasheed, J. A. Abdalla, and A. K. Al-Tamimi,
Behavior of reinforced concrete beams strengthened with externally
bonded hybrid fiber reinforced polymer systems, Mater. Des., vol. 53,
pp. 972–982, 2014. doi:10.1016/j.matdes.2013.07.087.
[4] A. Bilotta, et al., Bond efficiency of EBR and NSM FRP systems for
strengthening concrete members, J. Compos. Constr., vol. 15, no. 5,
pp. 757–772, 2011. doi:10.1061/(ASCE)CC.1943-5614.0000204.
[5] F. Ceroni, M. Pecce, A. Bilotta, and E. Nigro, Bond behavior of FRP
NSM systems in concrete elements, Compos. Part B Eng., vol. 43, no.
2, pp. 99–109, Mar. 2012. doi:10.1016/j.compositesb.2011.10.017.
[6] G. Wu, Z.-Q. Dong, Z.-S. Wu, and L.-W. Zhang, Performance and
parametric analysis of flexural strengthening for RC beams with
NSM-CFRP bars, J. Compos. Constr., vol. 18, no. 4, p. 4013051, 2013.
doi:10.1061/(ASCE)CC.1943-5614.0000451.
[7] A. Bilotta, F. Ceroni, E. Nigro, and M. Pecce, Efficiency of CFRP NSM
strips and EBR plates for flexural strengthening of RC beams and
loading pattern influence, Compos. Struct., vol. 124, pp. 163–175,
2015. doi:10.1016/j.compstruct.2014.12.046.
[8] L. De Lorenzis and J. G. Teng, Near-surface mounted FRP rein-
forcement: An emerging technique for strengthening structures,
Compos. Part B Eng., vol. 38, no. 2, pp. 119–143, Mar. 2007.
doi:10.1016/j.compositesb.2006.08.003.
[9] J. Teng, L. De Lorenzis, and B. Wang, Debonding failures of RC
beams strengthened with near surface mounted CFRP strips, J. Com-
pos. Constr., no. April, pp. 92–106, 2006. doi:10.1061/(ASCE)1090-
0268(2006)10:2(92).
[10] J. G. Teng, S. S. Zhang, and J. F. Chen, Strength model for end
cover separation failure in RC beams strengthened with near-surface
mounted (NSM) FRP strips, Eng. Struct., vol. 110, pp. 222–232, 2016.
doi:10.1016/j.engstruct.2015.11.049.
[11] S. B. Singh, A. L. Reddy, and C. P. Khatri, Experimental and
parametric investigation of response of NSM CFRP-strengthened
RC beams, J. Compos. Constr., vol. 18, no. 1, p. 4013021, 2014.
doi:10.1061/(ASCE)CC.1943-5614.0000411.
[12] G. M. Dalfré and J. A. O. Barros, Flexural strengthening of RC con-
tinuous slab strips using NSM CFRP laminates, Adv. Struct. Eng., vol.
14, no. 6, pp. 1223–1245, 2011. doi:10.1260/1369-4332.14.6.1223.
[13] R. A. Hawileh, Nonlinear finite element modeling of RC beams
strengthened with NSM FRP rods, Constr. Build. Mater., vol. 27, no.
1, pp. 461–471, Feb. 2012. doi:10.1016/j.conbuildmat.2011.07.018.
MECHANICS OF ADVANCED MATERIALS AND STRUCTURES 11
[14] T. K. Hassan and S. H. Rizkalla, Bond mechanism of near-surface-
mounted fiber-reinforced polymer bars for flexural strengthening
of concrete structures, ACI Struct. J., vol. 101, no. 6, pp. 830–839,
2004.
[15] T. Hassan and S. Rizkalla, Investigation of bond in concrete struc-
tures strengthened with near surface mounted carbon fiber reinforced
polymer strips, J. Compos. Constr., vol. 7, no. 3, pp. 248–257, 2003.
doi:10.1061/(ASCE)1090-0268(2003)7:3(248).
[16] I. A. Sharaky, R. M. Reda, M. Ghanem, M. H. Seleem, and H.
E. M. Sallam, Experimental and numerical study of RC beams
strengthened with bottom and side NSM GFRP bars having differ-
ent end conditions, Constr. Build. Mater., vol. 149, pp. 882–903, 2017.
doi:10.1016/j.conbuildmat.2017.05.192.
[17] M. A. Hosen, M. Z. Jumaat, and A. B. M. S. Islam, Side
near surface mounted (SNSM) technique for flexural enhance-
ment of RC beams, Mater. Des., vol. 83, pp. 587–597, 2015.
doi:10.1016/j.matdes.2015.06.035.
[18] A. A. Shukri, M. A. Hosen, R. Muhamad, and M. Z. Jumaat,
Behaviour of precracked RC beams strengthened using the side-
NSM technique, Constr. Build. Mater., vol. 123, pp. 617–626, 2016.
doi:10.1016/j.conbuildmat.2016.07.066.
[19] J. G. Teng, S. S. Zhang, and J. F. Chen, Strength model for end
cover separation failure in RC beams strengthened with near-surface
mounted (NSM) FRP strips, Eng. Struct., vol. 110, pp. 222–232, 2016.
doi:10.1016/j.engstruct.2015.11.049.
[20] S. S. Zhang and J. G. Teng, Finite element analysis of end cover sepa-
ration in RC beams strengthened in flexure with FRP, Eng. Struct.,
vol. 75, no. 6, pp. 550–560, 2014. doi:10.1016/j.engstruct.2014.06.
031.
[21] M. Rezazadeh, J. A. O. Barros, and H. Ramezansefat, End concrete
cover separation in RC structures strengthened in flexure with NSM
FRP: Analytical design approach, Eng. Struct., vol. 128, pp. 415–427,
2016. doi:10.1016/j.engstruct.2016.09.062.
[22] K. H. Mo, P. Visintin, U. J. Alengaram, and M. Z. Jumaat, Predic-
tion of the structural behaviour of oil palm shell lightweight con-
crete beams, Constr. Build. Mater., vol. 102, pp. 722–732, 2016.
doi:10.1016/j.conbuildmat.2015.10.184.
[23] D. Knight, P. Visintin, D. J. Oehlers, and M. S. Mohamed Ali, Short-
term partial-interaction behavior of RC beams with prestressed FRP
and steel, J. Compos. Constr., vol. 18, no. 1, p. 4013029, 2014.
doi:10.1061/(ASCE)CC.1943-5614.0000408.
[24] P. Visintin, D. J. Oehlers, C. Wu, and M. Haskett, A mechanics solu-
tion for hinges in RC beams with multiple cracks, Eng. Struct., vol.
36, pp. 61–69, Mar. 2012. doi:10.1016/j.engstruct.2011.11.028.
[25] A. A. Shukri, P. Visintin, D. J. Oehlers, and M. Z. Jumaat, Mechanics
model for simulating RC hinges under reversed cyclic loading, Mate-
rials (Basel)., vol. 9, no. 4, p. 305, 2016. doi:10.3390/ma9040305.
[26] D. Knight, P. Visintin, D. J. Oehlers, and M. S. Mohamed Ali,
Simulating RC beams with unbonded FRP and steel prestressing
tendons, Compos. Part B Eng., vol. 60, pp. 392–399, Apr. 2014.
doi:10.1016/j.compositesb.2013.12.039.
[27] P. Visintin, D. J. Oehlers, C. Wu, and M. C. Griffith, The reinforce-
ment contribution to the cyclic behaviour of reinforced concrete
beam hinges, Earthq. Eng. Struct. Dyn., vol. 41, pp. 1591–1608, 2012.
doi:10.1002/eqe.1189.
[28] P. Visintin, D. Oehlers, and M. Haskett, Partial-interaction time
dependent behaviour of reinforced concrete beams, Eng. Struct., vol.
49, pp. 408–420, Apr. 2013. doi:10.1016/j.engstruct.2012.11.025.
[29] P. Visintin, D. J. Oehlers, R. Muhamad, and C. Wu, Partial-interaction
short term serviceability deflection of RC beams, Eng. Struct., vol. 56,
pp. 993–1006, Nov. 2013. doi:10.1016/j.engstruct.2013.06.021.
[30] K. M. ud Darain et al., Strengthening of RC beams using
externally bonded reinforcement combined with near-surface
mounted technique, Polymers (Basel), vol. 8, no. 7, p. 261, 2016.
doi:10.3390/polym8070261.
[31] A. A. Shukri, K. M. ud Darain, and M. Z. Jumaat, The tension-
stiffening contribution of NSM CFRP to the behavior of strength-
ened RC beams, Materials (Basel), vol. 8, no. 7, pp. 4131–4146, 2015.
doi:10.3390/ma8074131.
[32] A. A. Shukri and M. Z. Jumaat, Simulating concrete cover sep-
aration in RC beams strengthened with near-surface mounted
12 A. A. SHUKRI ET AL.
reinforcements, Constr. Build. Mater., vol. 122, pp. 1–11, 2016.
doi:10.1016/j.conbuildmat.2016.06.048.
[33] D. J. Oehlers, P. Visintin, and W. Lucas, Flexural strength and ductil-
ity of FRP-plated RC beams: Fundamental mechanics incorporating
local and global IC debonding, J. Compos. Constr., vol. 20, no. 2, p.
4015046, 2015. doi:10.1061/(ASCE)CC.1943-5614.0000610.
[34] D. J. Oehlers, P. Visintin, M. Haskett, and W. M. Sebastian, Flexural
ductility fundamental mechanisms governing all RC members in par-
ticular FRP RC, Constr. Build. Mater., vol. 49, pp. 985–997, Dec. 2013.
doi:10.1016/j.conbuildmat.2013.02.018.
[35] R. Muhamad, M. S. Mohamed Ali, D. J. Oehlers, and M. Griffith,
The tension stiffening mechanism in reinforced concrete prisms, Adv.
Struct. Eng., vol. 15, no. 12, pp. 2053–2069, 2012. doi:10.1260/1369-
4332.15.12.2053.
[36] D. J. Oehlers, M. Haskett, M. S. Ali, W. Lucas, and R. Muhamad,
Our obsession with curvature in RC beam modelling, Adv. Struct.
Eng., vol. 14, no. 3, pp. 391–404, Jun. 2011. doi:10.1260/1369-
4332.14.3.391.
[37] A. K. Gupta and S. R. Maestrini, Tension stiffness model for rein-
forced concrete bars, J. Struct. Eng., vol. 116, no. 3, pp. 769–790, 1990.
doi:10.1061/(ASCE)0733-9445(1990)116:3(769).
[38] M. Haskett, D. J. Oehlers, and M. S. Mohamed Ali, Local and global
bond characteristics of steel reinforcing bars, Eng. Struct., vol. 30, no.
2, pp. 376–383, Feb. 2008. doi:10.1016/j.engstruct.2007.04.007.
[39] R. Muhamad, M. S. Mohamed Ali, D. Oehlers, and A. Hamid Sheikh,
Load-slip relationship of tension reinforcement in reinforced con-
crete members, Eng. Struct., vol. 33, no. 4, pp. 1098–1106, Apr. 2011.
doi:10.1016/j.engstruct.2010.12.022.
[40] R. Muhamad, D. J. Oehlers, and M. M. Ali, Discrete rotation deflec-
tion of RC beams at serviceability, Proc ICE Struct Build., vol. 164,
pp. 1–14, 2011.
[41] S. Popovics, A numerical approach to the complete stress-strain curve
of concrete, Cem. Concr. Res., vol. 3, no. 5, pp. 583–599, 1973.
doi:10.1016/0008-8846(73)90096-3.
[42] Y. Chen, P. Visintin, D. Oehlers, and U. J. Alengaram, Size-dependent
stress-strain model for unconfined concrete, J. Struct. Eng., vol. 140,
no. 4, p. 4013088, 2014. doi:10.1061/(ASCE)ST.1943-541X.0000869.
[43] D. Knight, P. Visintin, D. J. Oehlers, and M. S. Mohamed
Ali, Simulation of RC beams with mechanically fastened
FRP strips, Compos. Struct., vol. 114, pp. 99–106, Aug. 2014.
doi:10.1016/j.compstruct.2014.04.012.
[44] D. Oehlers, P. Visintin, T. Zhang, Y. Chen, and D. Knight, Flexural
rigidity of reinforced concrete members using a deformation based
analysis, Concr. Aust., vol. 38, no. 4, pp. 50–56, 2012.
[45] P. Achintha and C. Burgoyne, Moment-curvature and strain energy
of beams with external fiber-reinforced polymer reinforcement, ACI
Struct. J., vol. 106, no. 1, pp. 21–29, 2009.
[46] G. X. Guan and C. J. Burgoyne, Comparison of moment-curvature
models for fiber- reinforced polymer plate-end debonding studies
using global energy balance approach, ACI Struct. J., vol. 111, no. 1,
pp. 27–36, 2014.
[47] M. Achintha and C. Burgoyne, Fracture energy of the concrete-FRP
interface in strengthened beams, Eng. Fract. Mech., vol. 110, pp. 38–
51, Sep. 2013. doi:10.1016/j.engfracmech.2013.07.016.
[48] M. Achintha and C. J. Burgoyne, Fracture mechanics of plate debond-
ing: Validation against experiment, Constr. Build. Mater., vol. 25, no.
6, pp. 2961–2971, Jun. 2011. doi:10.1016/j.conbuildmat.2010.11.103.
[49] CEB-FIP, CEB-FIP Model Code 1990. London, UK: Thomas Telford
Ltd., 1993.
[50] L. De Lorenzis, Anchorage length of near-surface mounted fiber-
reinforced polymer rods for concrete strengthening – Analytical
modeling, ACI Struct. J., vol. 101, no. 3, pp. 375–386, 2004.